3T Siemens CMRR sLaser & Mega sLaser (for GABA) Basis Sets

Hello everyone,
I am a PhD student and relatively new to MRS and basis set generation. About 7 months ago, we set up MRS at our scanner with some collaborators for my study, and I am currently the sole user and responsible for it. I’ve encountered some technical issues getting the sequence to run on our Siemens Magnetom Skyra 3T scanner, but it seems to work fine now based on my limited experience and the pilot data. (Screenshot attached at the bottom of the post—would greatly appreciate it if someone could review it to see if the acquired data looks good).

Does anyone by chance have Basis Sets for the CMRR sLaser (TE/TR of 31ms/5000ms) and Mega sLaser sequences (TE/TR of 68.32ms/2000 ) (GABA) for a 3T Siemens Magnetom Skyra 3 Scanner or could point me in the direction of how to create one? I am mainly interested in Glu/GABA in V1 for both sequences.

I 've reached out to CMRR to see if they could provide a basis set for GABA-editing for the Mega-sLaser sequence, but they don’t have one available and told me that they would need experimental macromolecules obtained with that sequence and that the basis set depends on the specific pulses used in the acquisition. We had to increase the length of the excitation pulse of the sequence to be able to reach the required flip angle – not sure if that is relevant

Moreover, I tried generating a Mega-sLaser basis set using MRS cloud, but I am unsure if I did it correctly and if it is accurate. I used the default sequence values for edit on/edit off/pulse duration. When using this generated basis set (converted from LC Model to an osprey basis set) to analyze some pilot data I collected, Osprey runs into an error during fitting:

Index in position 3 exceeds array bounds. Index must not exceed 1.

Error in osp_fitMEGA (line 94)
basisSetDiff1.fids = basisSetDiff1.fids(:,:,3);

Error in OspreyFit (line 58)
[MRSCont] = osp_fitMEGA(MRSCont);

Error in osp_onFit (line 36)
MRSCont = OspreyFit(MRSCont);

Error while evaluating UIControl Callback.

Any help or guidance would be very much appreciated. Thank you in advance!

Voxel is in V1 in this pilot:

Hi Thomas,

Thanks for joining MRSHub. Here are a few thoughts:

  1. There are two ‘flavors’ of CMRR sequences, one by the Marjanska/Auerbach group, one by the Öz/Deelchand group. Which one is yours? They can be told apart by the sequence tag in the TWIX file header (there will be an eja string for the Auerbach and a dkd string for the Deelchand sequences).

  2. I believe the MRSCloud code simulates a relatively generic sLASER and does not differentiate the different CMRR sequences sufficiently. I have re-written the simulation code to be a bit more flexible, but it is not deployed in the cloud yet. I can probably help with the basis set depending on which CMRR flavor your sequence is.

  3. About your data: SNR and linewidth look OK, but I am concerned by the lack of co-edited Glx and the not-fully-saturated NAA in the edit-ON spectrum (indicating potential editing pulse frequency offset), a poorly resolved GABA signal with subtraction artefacts, as well as the lipid contamination (which might cause the subtraction artefact). For the lipids, you should probably move your voxel a few millimeters away from the scalp (do you have screenshots of your voxel placement?) or consider adding a lipid saturation band, although that’s a bit tricky with the skull curvature in the occipital area. My main concern is that you might have a systematic editing pulse frequency offset. Can you share the protocol PDF so we can inspect?

  4. There also seems to be considerable frequency drift, indicated by the slope of the frequency data points in the upper right plot. Are you running any fMRI/DWI scans before the MRS? It’s generally advised to run edited MRS on a ‘cold’ scanner, i.e. without sequences using heavy gradient duty cycles in the 30-60 minutes before.

1 Like

If you want something to crib off the MEGA sLASER sequence is detailed here: sequences/eja_svs_mslaser · master · William Clarke / WIN MRS Basis Sets · GitLab

Note that it’s only got the 7T editing pulse in for the time being. Both flavours of CMRR sLASER are in there as well.

1 Like

But also not too cold. We see a lot of drift in the first ~30 mins when switching the scanner on each morning.

Hi Georg,
Thank you so much for your detailed response and for the insights you’ve provided—it’s immensely helpful.

  1. I am using the ones by the Marjanska/Auerbach group. The TWIX file headers contain the “eja” string (slaser(eja_svs_slaser), Mega-sLaser(eja_svs_mslaser)). Prof. Marjanska has also guided us in adjusting the excitation pulses to fix technical difficulties we encountered on our scanner.

  2. It’s great to hear that you have developed a more flexible simulation code. I would be very interested in trying it out once it’s deployed. That would be awesome. I would appreciate any help you could provide with the basis set.

  3. Thank you for having a look at my data. Could you elaborate on what might cause an editing pulse frequency offset? We had to increase the excitation pulse duration from 2000 us to 2352 us and adjust the vapor 8 delay because increasing the pulse amplitude was not possible with our scanner configuration to achieve the required 90° flip angle. (Don’t know if this could play a role)

  • Attached is a screenshot of the voxel placement. Thank you for offering to inspect our protocol pdfs. I will export them from the scanner next Tuesday, including the recent changes, and share them with you.

  • Would it also be helpful to share the sequences without water suppression used as water reference files?

  1. Our scanner is heavily used for fmri scans (psychological research). The study before my pilot was running fmri scans. Between that and the pilot MRS scan was approximately a 20-30 min time window. I was just running a T1 for voxel placement and preparatory scans for the MRS scans beforehand. I could try running a pilot on a cold scanner the next time if that is helpful for evaluating the MRS data.
  • But due to the design of my study, which involves a baseline MRS for Glu/GABA, followed by a ~45-minute task with fMRI, and another MRS scan post-task running it on a ‘cold’ scanner may not be really feasible for the study itself. (The last 3 pilots included my whole study setup)

  • Would it still be advantageous to conduct the baseline MRS scan on a ‘cold’ scanner, even if the second MRS scan is approximately 15 minutes post-fMRI? How critical is it to maintain this condition for accurate comparative analysis between baseline and post-task MRS scans?

Thank you once again for your support and guidance. I look forward to your further suggestions.

Thank you, William! I really appreciate your help. I’ll definitely take a look, although I must admit, I might need some time to fully understand it— I’m still learning and not an MRI physicist. Thanks again for pointing me in the right direction!

Thank you! That’s very good to know, I’ll keep that in mind, when scheduling my scanner slots.

Hi @admin and @wclarke,

Attached is the protocol PDF for the sequences, along with the preparatory scans that were conducted beforehand. Thomas_MRS_study_longer_exc_MRS.pdf (112.6 KB)

The sequences are from the CMMR Spectroscopy package Release 2017-07 “Tokyo”
14 July 2017. The original msLaser protocol name is “eja_svs_mslaser_GOIA_32.”

Please let me know if there’s anything else you need for your review. Thank you in advance!